parca-dev / parca

Continuous profiling for analysis of CPU and memory usage, down to the line number and throughout time. Saving infrastructure cost, improving performance, and increasing reliability.
https://parca.dev/
Apache License 2.0
4.03k stars 210 forks source link

Callgraph visualization #304

Open brancz opened 2 years ago

brancz commented 2 years ago

Icicle graphs are the most popular way to explore performance data, but callgraphs can also be a very useful visualization when different stack traces call the same function for example.

metalmatze commented 2 years ago

As a workaround (until we decide to put up with graphviz or find an alternative) we should make pprof profiles downloadable so that people can use go tool pprof profile.pb.gz and get that functionality.

brancz commented 2 years ago

We should do that regardless!

metalmatze commented 2 years ago

Example screenshot graph

syepes commented 2 years ago

@brancz Will this feature be added back to parca? We are still running some conprof instances just to have this kind of builtin visualization, as it simplifies the debugging

metalmatze commented 2 years ago

@monicawoj is actually actively investigating this! :tada: https://github.com/parca-dev/parca/pull/1132